BACKGROUND Bleeding in the first trimester of pregnancy is a common phenomenon, associated with early pregnancy loss. In many instances a subchorionic hematoma is found sonographically. OBJECTIVE To evaluate the possible benefit of bed-rest in women with threatened abortion and sonographically proven subchorionic hematoma, and to examine the possible relationship of duration of vaginal bleeding, hematoma size, and gestational age at diagnosis to pregnancy outcome. METHODS The study group consisted of 230 women of 2,556 (9%) referred for ultrasound examination because of vaginal bleeding in the first half of pregnancy, who were found to have a subchorionic hematoma in the presence of a singleton live embryo or fetus. All patients were advised bed-rest at home; 200 adhered to this recommendation for the duration of vaginal bleeding (group 1) and 30 continued their usual lifestyle (group 2). All were followed with repeated sonograms at 7 day intervals until bleeding ceased, the subchorionic hematoma disappeared, or abortion occurred. The groups were compared for size of hematoma, duration of bleeding, and gestational age at diagnosis in relation to pregnancy outcome (spontaneous abortion, term or preterm delivery). RESULTS The first bleeding episode occurred at 12.6 +/- 3.4 weeks of gestation (range 7-20 weeks) and lasted for 28.8 +/- 19.1 days (range 4-72 days). The women who adhered to bed-rest had fewer spontaneous abortions (9.9% vs. 23.3%, P = 0.006) and a higher rate of term pregnancy (89 vs. 70%, P = 0.004) than those who did not. There was no association between duration of vaginal bleeding, hematoma size, or gestational age at diagnosis of subchorionic hematoma and pregnancy outcome. CONCLUSIONS Fewer spontaneous abortions and a higher rate of term pregnancy were noted in the bed-rest group. However, the lack of randomization and retrospective design of the outcome data collection preclude a definite conclusion. A large prospective randomized study is required to confirm whether bed-rest has a real therapeutic effect.